SimPA: A Sentence-Level Simplification Corpus for Public Administration

A sentence-level simplification corpus contains 1,100 original sentences from the Public Administration domain. The corpus includes 3,300 lexical simplifications and 1,100 syntactic simplifications, collected manually through a two-stage annotation process. It was created by Carolina Scarton at the University of Sheffield for the EU H2020 SIMPATICO project.

Use Cases

• Training text simplification models based on the manually simplified sentence pairs.
• Evaluating lexical simplification approaches based on the multi-reference (three per sentence) annotations.
• Analyzing syntactic simplification techniques based on the isolated syntactic-level data.
• Researching domain adaptation for NLP in public administration based on the domain-specific content.

Strengths

• Provides 1,100 original sentences with manual simplifications.
• Offers 3,300 lexically simplified versions, enabling multi-reference evaluation.
• Separates lexical and syntactic simplifications, allowing for isolated analysis.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.
• The final aim is over 10K sentences, suggesting the current 1,100-sentence version is a partial collection.

Provenance

Source

University of Sheffield, EU H2020 SIMPATICO project.

Collection Method

Manual simplification via a two-stage annotation process (lexical then syntactic).

Time Range

null

Freshness

Last updated is unknown.

Geography

null